1
Oxygen Mass Transfer in the Albion ProcessTM: from the Laboratory to the Plant
Paul Voigt, Mike Hourn and Daniel Mallah
Glencore Technology, Australia
ABSTRACT
The successful commissioning and ramp up of the Albion ProcessTM at the GPM Gold Project relied
on the successful scaling up of the process from batch and continuous pilot plant campaigns (Voigt,
2016) Critical information about reaction kinetics and residence time, grind size and pulp density
were determined at the laboratory scale and successfully applied to the commercial scale. A
limitation of small scale testwork, is that some parameters cannot be measured reliably and scaling
up is a function of the physical size of the equipment which isn’t possible to test with laboratory
scale equipment. Oxygen mass transfer rate is one such parameter since this is a complex
interaction of many factors including slurry temperature, solution and slurry chemistry, slurry
viscosity, agitator type, dimensions and power, oxygen bubble residence time, oxygen purity, tank
geometry and oxygen injection technique. Oxygen generation represents an important operating
cost for the Albion ProcessTM. Pivotal to the Albion ProcessTM operating economically at
atmospheric pressure is the capability to efficiently transfer oxygen while utilising as much oxygen
injected to the process as possible. To respond to this Glencore Technology developed the
HyperSpargeTM supersonic gas injector. This paper compares the HyperSpargeTM against other
sparging techniques to quantify the benefits of oxygen injection via a supersonic gas jet on scale up
of the oxygen mass transfer system. The paper then examines plant survey data from the GPM
Gold Project to demonstrate the very high oxygen utilisation that can be achieved with a correctly
designed oxygen mass transfer system.
2
INTRODUCTION
One of the main challenges in chemical reactor design is the scale-up of processes from the
laboratory to industrial scale. One particular challenge are those processes which cannot be
faithfully simulated with an experiment at the laboratory due to the physical dimensions and
complexity of the real system. One example of this is the design of oxygen mass transfer systems in
atmospheric oxidative leaching such as the Albion ProcessTM. Some of the problems with the small
size of equipment are the specific agitator power input is artificially high, the bubble residence time
in the vessel is artificially low, the oxygen partial pressure at the base of the vessel is artificially low
and the way oxygen is injected into the process may not be the same as the industrial process. Some
of the problems of the complexity of the system is the dynamic nature of the process including the
presence of solids, recycle streams and minor elements plus the variances within the reality of a
process plant such as disruptions or variance of feed quality.
It is critical that for such processes the fundamentals are understood so that key data can be
collected from appropriately designed experiments at the laboratory scale for input into a proven
design approach to render the industrial scale process effective and fit-for-purpose.
To maximise oxygen mass transfer efficiency, operational availability, simplicity and safety,
Glencore Technology (GT) developed the HyperSpargeTM supersonic gas injector. Originally for
injecting oxygen to the Albion ProcessTM, the HyperSpargeTM has found application in other
processes using air, oxygen, sulphur dioxide and gas mixtures. Over 400 HyperSpargeTM units are
installed globally in duties from in-line slurry conditioning, fermentation, waste-water treatment,
CIL/CIP processes and oxidative leaching.
This paper compares the HyperSpargeTM to other gas injection techniques such as ring spargers and
converging nozzles. The key data are identified for scale-up from the laboratory to industrial scale.
Survey data are examined from the GPM Gold plant in Armenia where the oxygen mass transfer
system was successfully designed and comments are made about the operational advantages of the
HyperSpargeTM compared to other systems such as sintered spargers.
OXYGEN MASS TRANSFER
The Albion ProcessTM is an atmospheric oxidative leaching process developed by GT in 1994 and is
described extensively in the literature (Hourn & Turner, 2010; Hourn & Turner, 2012; Hourn et al.,
2014; Voigt et al., 2015; Senshenko et al., 2016). Oxidation of pyrite is one of the key reactions in the
Albion ProcessTM chemistry as shown in Equation 1.
FeS2 + 15/4O2 + ½ H2O = ½ Fe2(SO4)3 + ½ H2SO4 [1]
The first step of the oxidation mechanism is the dissolution of oxygen in the liquid phase since
electron transfer occurs in part through the action of the ferric and ferrous ion couple and the action
of dissolved oxygen reacting directly with pyrite. Hence the rate at which oxygen is transferred is
critical to the process efficiency and can be simplified for a reactor as shown in Equation 2
(Middleton & Smith, 2004).
O2 Transfer Rate = kLa.V.(Csat – C) [2]
3
Where kL is the liquid film transfer co-efficient (m.s-1), a is the specific gas surface area (m-1), V is
reactor volume (m3), Csat is the oxygen solubility at saturation (g.m-3) and C is the steady state
oxygen level (g.m-3). The terms kL and a are normally combined to represent the oxygen mass
transfer co-efficient which is the parameter to be maximised.
For practical purposes the kLa needs to be maximised by determination of a relationship with
power input through the sparger and the agitator and then selection of equipment that satisfies the
oxygen transfer rate requirements of the system. Middleton (1992) simplified the equation relating
kLa to power input is shown in Equation 3.
kLa = K . (Us)a . (Pg/V)b [3]
Where K is a coalescence constant, Us is superficial gas velocity (m.s-1) and Pg is absorbed power
(W). A correction is also applied for temperature since increasing temperature will reduce slurry
viscosity and surface tension increasing the mass transfer interfacial area. The coefficients a and b
are empirical constants that are system specific and are determined through testwork.
There are many theoretical correlations between kLa and energy input reported in the literature and
the majority tend to be based on data in biochemical reactors meaning the extension to oxidative
leaching can be problematic with limited applicability because the agitation intensity, system
chemistry, operating temperature and oxygen injection methods are very different (Van’t Reit, 1979;
Vasconcelos et al., 1998; Oguz et al., 1987; Moo-Young & Blanch, 1981). Most of these correlations
are also derived at very small scale, with laboratory vessels of 10 litres or less in size.
The best way to achieve successful industrial scale up is to determine the a and b empirical
constants for equation 3 for a given system through specific experiments that reflect the actual
system conditions. Once determined for a specific system, Equation 3 is then used along with other
reactor design equations and experience to calculate the tank geometry and power input required
from the agitator and gas injector to achieve the required kLa. GT has developed an experimental
procedure to determine the a and b empirical constants for a given system which will result in a
successful scale up with minimal error. An example of using this method to compare the
experimental and calculated kLa for a copper leach system is shown in Figure 1.
Figure 1 Experimental and calculated kLa for a system using GT experimental methodology
4
INJECTION OF OXYGEN
Importance of oxygen injection method
The method in which oxygen is injected has a large bearing on the efficiency of the oxygen mass
transfer. Referring to Equation 2, the liquid film co-efficient (kL) is proportional to the oxygen
diffusion coefficient and inversely proportional to the liquid film thickness. The consequence for
oxygen injector selection is that high pressure gas injection maximises shear and significantly
erodes the liquid film thickness resulting in increasing kL. Referring to Equation 3, the surface area
(a) is proportional to power absorbed into the system relative to system volume and superficial gas
velocity. Surface area is also inversely proportional to bubble size. The consequence for oxygen
injector selection is to maximise shear and minimise bubble size but also to select an agitator that
provides sufficient complimentary power input.
Options for oxygen injection
There are many gas injectors available on the market. The converging nozzle type and ring spargers
(essentially open pipes) and among the most common from an operational perspective. Sintered
spargers find application in flotation machines and chemical reaction systems and are well suited to
the laboratory but are plagued with operational issues at the industrial scale, particularly in
processes with chemical reactions due to the fast accumulation of reaction products and the
continual need to clean them.
GT performed a number of experiments comparing the performance of an open pipe sparger, a
converging nozzle type sparger and a newly developed supersonic sparger by GT, the
HyperSpargeTM.
Open pipes and sparge rings do not transfer a significant amount of energy into the leach vessel,
with all the energy required for oxygen mass transfer provided by the impeller. For these spargers,
the gas is compressed to just above the hydrostatic head level in the slurry tank and the pressure
drop across the tip of the injection point is low, resulting in low gas speeds.
Convergent nozzles, such as the CPT Slamjet and Minnovex gas spargers, were developed for
flotation column aeration duties, and have found some use in slurry oxidation systems. These
nozzles converge to a narrow opening, and the gas is compressed to several atmospheres above the
hydrostatic head level in the slurry tank. The pressure drop across the tip of the injection point is 3
– 5 atmospheres, resulting in higher gas speeds at the point of injection. This results in more energy
being put into the gas from the sparging system, and so less energy is required at the impeller.
Testwork carried out by GT has found that for oxygen mass transfer purposes, energy input to the
system via the gas sparging system is more efficient than energy input via the impeller, as long as
the majority of the energy released from the gas pressure drop across the sparger is converted to
kinetic energy and not lost as heat. As such GT developed a supersonic gas injector, the
HyperSpargeTM, to further improve the efficiency of the gas injection system. The HyperSpargeTM
used a converging-diverging nozzle to maximise energy recovery from the gas pressure drop by
accelerating the gas to supersonic velocities.
5
Efficiency of oxygen injection methods
Energy is put into the gas sparging system by pressurising the gas prior to the sparger, to store
energy in the gas stream. The theoretical energy input to compress the gas is defined in Equation 4.
W = (n/(n-1))mRT[(P2/P1)(n-1)/n – 1 [4]
Where n is isothermal efficiency, W is work (J), P1 is outlet pressure (N.m-2) and P2 is inlet pressure
(N.m-2), T is temperature (K), R is the ideal gas constant (8.314472 J.K-1.mol-1) and m is moles.
The pressure energy is then recovered as work into the system, ideally the efficiency of this
conversion is maximised. High speed gas injection maximises turbulence at the injection point,
which reduces the liquid film boundary layer around the gas bubble, increasing the liquid film
diffusion rate, and so increasing the kL value. The higher kinetic energy of the gas will translate to a
higher interfacial area, which also assists mass transfer.
In a conventional pipe or ring sparger, the nozzle diameter is relatively large, resulting in a low
velocity gas flow, which remains constant along the length of the nozzle. In this configuration,
higher gas flows can lead to larger bubble sizes and consequently smaller interfacial areas (a). As a
general rule, the size of the bubble generated from a conventional sparger will be close to the size of
the orifice.
The formation of fine bubbles requires a specific type of nozzle to be fitted to the end of the sparger.
Nozzles can normally be described as convergent (narrowing down from a wide diameter to a
smaller diameter in the direction of the flow) or divergent (expanding from a smaller diameter to a
larger one in the direction of the flow).
Convergent nozzles accelerate subsonic gases, with the dynamic pressure decreasing from the
nozzle opening to the narrowest point as the gas is accelerated. Once the ratio of the gas pressure at
the nozzle throat to the feed pressure reaches a critical value, the flow will reach sonic velocity. This
forms a shock wave in the nozzle throat. The nozzle is then said to be choked, and the gas flow is at
Mach 1. Gas flow cannot be accelerated beyond Mach 1 in the throat of the nozzle.
Increasing the nozzle feed pressure further will not increase the velocity of the gas at the throat of
the nozzle, and so any additional pressure energy is then stored in the gas in the form of heat.
Divergent nozzles decelerate subsonic gases; the gas expands in the nozzle and slows down.
However, when the gas enters the divergent section at Mach 1, the divergent nozzle will accelerate
the gas, as the stored heat energy is converted to velocity. Combination convergent-divergent
nozzles, such as those used in the HyperSpargeTM can, therefore, accelerate gases that have choked
in the convergent section to supersonic speeds. These nozzles develop thrust by converting the
stored heat that builds up in the gas in the throat into velocity when in the divergent section of the
nozzle.
When gas is accelerated to velocities above the speed of sound in the divergent section of the
nozzle, the local pressure drops further, and does not recover to ambient pressure until the gas
passes through the resulting shock wave as illustrated in Figure 2.
6
Figure 2 Collapsing bubbles crossing the shock wave of a supersonic nozzle (Kawamura et al., 2004)
The low pressure that is generated in the divergent section of the nozzle leads to collapse of the gas
bubble into a fine mist, and results in a dramatic increase in the gas surface area. This is shown in
the figure above, where the gas bubble leaves the throat of the nozzle at a size that is fairly similar
to the size of the throat diameter, and then collapses to a fine mist of micro-bubbles in the low
pressure zone in the divergent section. This collapse is triggered by the gas crossing the shock wave
that is set up once the gas is choked in the throat of the nozzle. This shock wave can extend
anywhere from the nozzle throat to the end of the nozzle and beyond, depending on the pressure
applied to the nozzle.
A convergent-divergent nozzle has two important roles. The design of the nozzle determines the
exit velocity for a given pressure and temperature. And because of flow choking in the throat of the
nozzle, the nozzle design also sets the mass flow rate through the nozzle. Therefore, the nozzle
design determines the thrust of the gas injection system. By changing the shape of the nozzle and
the flow conditions upstream and downstream, you can control both the amount of gas that passes
through the nozzle and the thrust generated by the nozzle. A convergent-divergent nozzle is
represented below.
Figure 3 Cross section of a convergent-divergent nozzle (NASA, 2015)
EXPERIMENTAL
A number of experiments were performed to compare the nozzle types evaluating on the variables
of thrust and oxygen mass transfer co-efficient (kLa). Experiments were performed in a jacketed,
1750 litre tank fitted with four baffles and a lid to minimise evaporative loss. The process solution
employed was a copper leach solution at pH 1, with 15 g/l-1 copper. High purity gaseous oxygen
and nitrogen was used. The nozzles employed for the experiments were:
10mm open pipe
Converging high pressure sparger, 4mm exit diameter
7
Converging-diverging HyperSpargeTM nozzle, Mach 3 (MV), 1mm throat and 5mm exit
diameter
Converging-diverging HyperSpargeTM nozzle, Mach 1 (MT), 1mm throat and 1.7mm exit
diameter
Thrust
Thrust is a direct measure of the efficiency of conversion of pressure energy in the incoming gas
stream into momentum. The higher the thrust measured, the more efficient the nozzle will be in
converting pressure energy into mass transfer without loss to heat dissipation. The nozzles were
positioned perpendicular and centred to a balance surface. The gas plume impinged the balance’s
surface and the mass imparted recorded. For an identical energy input in the sparging gas (ie fixed
flow and delivery pressure), each nozzle generated a different degree of thrust. The data are shown
in Figure 4.
Figure 4 Developed thrust measured at 200kPa
The results above show that at low oxygen flow all of the profiled nozzles exhibited similar thrust
values; less than 3 grams (g) of thrust recorded by the scale. However, at higher flows, the nozzles
have a very different behaviour. The open pipe sparger did not develop any thrust at all across the
entire range of gas flowrates, and so did not exert any force on the balance. The convergent nozzle
showed an almost linear response, with thrust increasing with flowrate, up to a maximum of 5 g at
13 lpm of gas flow.
Both GT HyperSpargeTM supersonic nozzles generated significantly more thrust than the open pipe
or convergent nozzle. In general, the nozzle designed to produce maximum thrust outperformed
the high velocity nozzle across all flow rates, achieving a thrust of 18 g at a gas flowrate of 13 lpm.
This compared well to the maximum theoretical thrust for a nozzle discharging 13 lpm of gas flow
of 21 g. This meant that 85 % of the pressure energy in the gas stream was converted to momentum
8
for the HyperSpargeTM MT nozzle profile. Both supersonic nozzles also displayed a considerable
steepening of the thrust curve at gas flowrates above 11 lpm, close to the point where the nozzles
were choked, indicating superior thrust as the gas speeds exiting the convergent-divergent nozzles
approached supersonic speeds.
Oxygen mass transfer co-efficient (kLa)
For temperatures up to 50oC oxygen mass transfer co-efficient was determined using the Dynamic
Gassing Out Method (Taguchi and Humphrey, 1966). Above 50oC the combined sulphite method
was used (Puskeiler, 2005). Experiments were conducted with and without solids in the reactor. The
results are shown in Figure 5.
The GT MV and MT nozzle designs resulted in a consistently higher kLa value across the entire
range of gas flow rates tested relative to the open pipe and the convergent nozzle. As the gas flow
rate approached choked flow in the GT HyperSpargeTM MV and MT nozzles, the difference in kLa
was more pronounced, and was up to 300 % of the value observed for the conventional open pipe
and 50% compared to a converged nozzle.
Figure 5 kLa at different supply pressure
The GT HyperSpargeTM MV and MT nozzles showed similar behaviours because both are
convergent-divergent type nozzles. Of the two HyperSpargeTM nozzles, the MT nozzle provided the
highest kLa, gas utilisation and thrust values, typically by 25 % relative to the MV nozzle at a set gas
flow rate. The higher thrust generated in the MT nozzle at the expense of velocity explains this
difference.
9
SCALE UP TO INDUSTRIAL SCALE
Through the scientific process, GT established that supersonic gas injection was technically a
superior way to inject gas into a slurry requiring oxygen. The next step was to develop the
converging-diverging nozzle into a product that can be used at the industrial scale.
The HyperSpargeTM
The HyperSpargeTM was developed by GT as a complete gas injection solution. At the heart of the
device is a converging-diverging nozzle fabricated from a specially selected wear-resistant ceramic
which accelerates gases to supersonic velocities for injection into solutions and slurries. The nozzle
is designed to minimise pressure energy loss. A schematic of the HyperSpargeTM and close up of the
nozzle is shown in Figure 6.
Figure 6 HyperSpargeTM supersonic gas injector and nozzle
Apart from the advantages of superior oxygen mass transfer, the HyperSpargeTM is safe and easy to
use. The HyperSpargeTM can operate over a range of pressures which can be adjusted to regulate
flow. The nozzles can be easily interchanged with the threaded nozzle holder allowing the use of
different nozzle diameters and hence flow operating envelopes. One attractive feature is that the
sparger can be inserted and removed from a vessel while it is full of slurry spillage free without
using any additional removal device plus with a throat diameter of typically 7 mm, the sparger
does not become blocked and self-cleans by the action of the gas flow. This means less operational
downtime and process interruption.
10
Oxygen Mass Transfer System Design
During the GPM Gold Albion ProcessTM plant development, key oxygen mass transfer experiments
were conducted to determine the kLa under conditions as close as practicable for temperature and
slurry conditions. The data was used by GT to then correctly size the gas injectors, agitators and
select ideal tank dimensions. The design criteria for the GPM Gold Plant is shown in Table 1.
Table 1 GPM Gold Albion ProcessTM Plant Design Criteria
Parameter Units Nominal Design
Throughput tph 13.1 14.5
S2- concentration % 17.6 20.0
S2- oxidation % 76 76
Oxygen Utilisation % 90 80
Required kLa m.s-1 0.12 0.14
CIL Gold recovery % 92 90
The economics of the industrial process rely on the efficiency of oxygen transfer to the system to
maximise transfer rate (achieving design residence time) and maximise transfer efficiency
(utilisation of the oxygen that is injected.
A survey was completed to collect plant data and assess the transfer rate and transfer efficiency.
Since an industrial plant was the subject of the survey, the plant couldn’t be taken off-line to
replicate laboratory experiments. Instead, the oxidation extent and oxygen consumption from the
operating plant were compared against the design. The conditions of the survey were a throughput
rate of 14.0 tph concentrate and a sulphide concentration of 17.0%.The survey data for oxidation
extent and gold recovery are shown in Figure 7.
11
Figure 7 Sulphide Oxidation survey of GPM Gold Albion ProcessTM Plant.
Figure 7 shows that at the design throughput the sulphide oxidation is achieved along with
corresponding gold recovery.
The oxygen utilisation was determined by comparing the oxygen injection to the process during the
survey period against the modelled values. The results are shown in Figure 8.
Figure 8 Oxygen utilisation at the GPM Gold Albion ProcessTM Plant
12
Figure 8 shows that the cumulative oxygen utilisation is in excess of the design value of 90%
achieving 93%. This observation is consistent with other multi-tank Albion ProcessTM plants where
it appears oxygen that is dissolved in the first few tanks of the leaching train is utilised later in the
process.
Both Figures 7 and 8 illustrate that the oxygen mass transfer system was correctly and successfully
designed.
CONCLUSIONS
Oxygen injection using convergent-divergent nozzles generate superior thrust and oxygen mass
transfer compared to other gas injection techniques. Power delivered to the system is more efficient
through gas injection rather than mechanical agitation. The HyperSpargeTM is a development of the
convergent-divergent nozzle and offers additional advantages over other sparging technologies
contributing to a safer work environment, maximising process run-time and optimising energy
input through the agitator. GT successfully scaled up the oxygen mass transfer system from the
laboratory to the industrial scale at the GPM Gold Albion ProcessTM plant which achieves greater
than design performance in terms of oxygen mass transfer and oxygen utilisation.
REFERENCES
Hourn, M., & Turner, D.W. (2010). Albion Process for treatment of refractory ores, Proceedings of
ALTA Conference 2010, Perth, Australia.
Hourn, M., & Turner, D.W. (2012). Commercialisation of the Albion Process, Proceedings of ALTA
Conference 2012, Perth, Australia.
Hourn, M., Voigt, P., & Turner, D.W. (2014). Development of the Albion Process plant to treat
refractory concentrates from the GPM Gold Project, Proceedings of Hydroprocess Conference, Vina del
Mar, Chile.
Kawamura, T., Fujiwara A., Takahashi, T., Kato, H., Matsumoto, Y. & Kodama, Y. (2004). The effect
of the bubble size, dispersion and skin friction reduction. Proceedings of On smart control of
turbulence, Tokyo, Japan.
Middleton, J.C. (1992). Gas-liquid dispersion and mixing. In N. Harnby, M.F. Edwards, A.W.
Nienow (Ed.), Mixing in the Process Industries (pp. 322–363).
Middleton, J.C. & Smith, J.M. (1992). Gas-liquid mixing in turbulent systems. In E.L. Paul, V.
Atiemo-Obeng, S.M. Wiley (Ed.), Handbook of industrial mixing: Science and practice (pp. 585–638).
Moo-Young, M. & Blanch, H.W. (1987). Design of biochemical reactors: Mass transfer criteria for
simple and complex systems, Adv. Biochem. Eng., 19, 1–69.
NASA (2015). Nozzle design – converging/diverging (CD) nozzle,
https://www.grc.nasa.gov/www/k-12/airplane/nozzled.html, retrieved January 2017.
Oguz, H., Brehm, A. & Deckwer, W.D. (1987). Gas/liquid mass transfer in sparged agitated slurries,
Chemical Engineering Science, 42, 1815–1822.
Puskeiler, R. & Weuster-Botz, D. (2005). Combined sulfite method for the measurement of the
oxygen transfer coefficient kLa in bioreactors, Journal of Biotechnology, 120(4), 430–438.
13
Senshenko, A.Y., Aksenov A.V., Vasiliev A.A., & Seredkin, Y.G. (2016). Technology for processing
of refractory gold-containing concentrates based on ultrafine grinding and atmospheric oxidation,
Proceedings of IMPC Conference, Montreal, Canada.
Taguchi, H. & Humphrey, A.E. (1966). Dynamic measurement of volumetric O2 transfer coefficient
in fermentation systems, Journal of Fermentation Technology, 44, 881–889.
Van’t Riet, K. (1979). Review of measuring methods and nonviscous gas-liquid mass transfer in
stirred vessels, Ind Eng Chem Process Des Dev, 18, 357–364.
Vasconcelos, J.M.T., Alves, S.S., Nienow, A.W. & Bujalski, W. (1998). Scale-up of mixed gassed
multi-turbine agitate vessels, Canadian Journal of Chemical Engineering, 76, 398–403.
Voigt, P., Hourn, M., Mallah, D. & Turner, D.W. (2015). Commissioning and ramp up of the Albion
Process at the GPM Gold Project, Proceedings of World Gold Conference, Johannesburg, South Africa.
Voigt, P., Hourn, M. & Mallah, D. (2016). Treatment of low grade materials, Proceedings of MINEX
Conference, Moscow, Russian Federation.
Top Related